Study design
This prospective study was conducted to evaluate the utility of a large GS kit to obtain more tumor cells in patients with NSCLC. We prospectively enrolled patients scheduled to undergo EBUS-GS TBB for peripheral lung lesions suspected of being NSCLC and performed EBUS-GS TBB using large GS kit. Furthermore, we enrolled consecutive patients with NSCLC in whom tissue samples were previously obtained by EBUS-GS TBB using a small GS kit. We compared tumor cell numbers of samples obtained from patients who were histologically diagnosed with NSCLC by TBB with large GS (prospective large GS group) and those obtained from patients who were previously diagnosed with NSCLC by TBB with small GS (small GS cohort).
Patients
The eligibility criteria were age ≥20 years and patients with undiagnosed peripheral lung lesion, suspected of being NSCLC on computed tomography (CT). Written informed consent was obtained from all patients. The peripheral pulmonary lesion was defined as not visible by bronchoscopy. Exclusion criteria were patients with visible lesion on bronchoscopy, those who underwent re-biopsy after treatment of lung cancer, those who had severe comorbidities such as severe cardiac diseases or insufficient pulmonary function, patients who could not discontinue antiplatelet or anticoagulant therapy, pregnant woman, and patients judged to be unsuited for this study by their physicians. We also enrolled consecutive NSCLC patients who previously underwent EBUS-GS TBB using a small GS kit in our institution. We applied the opt-out method to obtain consent on retrospectively enrolled patients. This study was approved by the ethics committee of Kobe University (300016) and was conducted in accordance with the Helsinki declaration. This study is registered in the University Medical Hospital Information Network in Japan (UMIN 000032599, https://upload.umin.ac.jp/cgi-open-bin/ctr/ctr_view.cgi?recptno=R000036780).
CT evaluation of the lesion
Two pulmonologists evaluated the lesion size, location, presence, or absence of bronchus sign [11]. The location of the lesion was classified as central, intermediate, or peripheral based on the distance from the hilum on CT. Lesions located within the inner third area were considered central, those located within the middle third area were intermediate, and those located within the outer third area were considered peripheral [12]. The lesion characteristics on CT scan were classified as solid, part-solid ground-glass opacity (GGO), and pure GGO. Subsegmental bronchi were regarded as third-generation bronchi, and the number of bronchi was calculated by adding the number of further branchings.
Bronchoscopy procedure
All patients underwent CT scanning (slice width: 1.0 mm) before bronchoscopy. The bronchial path to the lesion was planned using a chest CT or virtual bronchoscopic navigation software (Bf-NAVI®; Cybernet Systems, Tokyo, Japan) that automatically created a virtual bronchial image of the target lesion [13]. Bronchoscopy was performed through the oral route under local anesthesia with conscious sedation. All bronchoscopies and devices were manufactured by Olympus, Tokyo, Japan. The bronchoscope, BF 1T260 (5.9 mm outer diameter, 2.8 mm working channel diameter) or 1TQ290 (5.9 mm outer diameter, 3.0 mm working channel diameter), was advanced to the target lesion through the planned bronchial route. A curette-type inductor (CC-6DR-1) was permitted to use to insert the GS into the appropriate bronchial branch. When the bronchoscope reached near the target lesion, the large GS (SG-201C or SG-401C, 2.5 mm external diameter) was inserted with UM-S20-20R radial EBUS probe through the working channel. If the target lesion could be visualized under EBUS, the EBUS probe was withdrawn and the forceps (FB-231D, diameter: 1.9 mm) were inserted through the GS. Specimens were obtained from the same lesion using forceps introduced into the GS as well as by brushing twice. After at least five specimens were obtained, the GS was aspirated with 20 mL of negative pressure for 20 seconds to collect cells and withdraw the GS [14]. If the target lesion was not visible by EBUS, the bronchoscope was changed to the thinner type—BF P290 (4.2 mm outer diameter, 2.0 mm working channel diameter) or BF P260F (4.0 mm outer diameter, 2.0 mm working channel diameter)—and advanced to the target lesion. When the thinner bronchoscope reached the lesion, the small GS (SG-200C or SG-400C, 1.9 mm external diameter) was inserted with UM-S20-17S radial EBUS probe through the working channel. Similar to the procedure with large GS TBB, specimens were obtained using forceps (FB-233D, diameter: 1.5 mm) and a brush introduced into the GS. Additional conventional TBB using biopsy forceps (FB-231D) and transbronchial needle aspiration using aspiration needle (MAJ-64) were permitted. X-ray fluoroscopy was intermittently used to guide the EBUS probe and biopsy devices to the target lesion and confirm movement of the devices during sample collection. We recorded the number of branches that can be observed under bronchoscopy and endobronchial ultrasonography images.
Pathological evaluation
Pathological diagnosis was confirmed by a histopathologist based on hematoxylin-eosin (HE) cell staining. After a confirmed diagnosis of NSCLC, we evaluated the number of tumor cells and sample size of the first consecutive five specimens in the large GS group and small GS cohort. We scanned the HE-stained slides with a scanner (Nano zoomer® 2.0RS, HAMAMATSU, Japan). A pathologist and a cytologist, blinded to the clinical details, manually counted the number of tumor cells and evaluated the proportion of tumor cells in all nuclear cells using imaging software (NDP scan® 2.5, HAMAMATSU, Japan), and average number and proportion assessed by the two evaluators. The sample size was measured by the cumulative area of individual specimens using imaging software (cellSens® standard, Olympus, Japan). PD-L1 was evaluated using the samples containing the largest number of tumor cells obtained by GS-TBB.
Endpoints
The primary endpoint was comparison of the mean tumor cell number per glass slide of first five consecutive samples between the prospective large GS group and the small GS cohort. The secondary endpoint was to evaluate the number of tumor cells containing the largest number of tumor cells among first five consecutive samples, sample size, proportion of tumor cells in nucleated cells, and the success rate of PD-L1 testing between prospective large GS group and small GS cohort. In the prospectively enrolled patients, we evaluated factors associated with change to a thinner bronchoscope in order to clarify the characteristics of lesions that could be obtained by TBB with large GS. We evaluated the tumor diameter, tumor location, tumor characteristics, CT bronchus sign, bronchus generation constructed by VB images, and the bronchus generation visible by bronchoscopy between lesions that did not require to change to the thinner bronchoscope (not changed group) and those that needed this change (changed group). Last, we evaluated the safety of prospectively performed bronchoscopy.
Statistical analyses
In a pilot study, the average tumor cell counts in two patients who underwent TBB with large GS were 1410 and 302. Further, in three cases that underwent EBUS-GS TBB using small GS, the average tumor cell counts were 192, 407, and 675. The mean difference between the two groups was 431 (SD: 855). The sample size was calculated assuming that the mean difference between the two groups was 431, with an alpha level set at 0.2 (two-sided) and detection power of 80%. The minimum sample size was calculated as 36 for each group (prospective large GS group and small GS cohort), and was set at 80 for the prospective large GS group assuming that cases that required to change the thinner bronchoscope, whose pathological diagnosis other than NSCLC and deviation. The chi-square or Fisher’s exact tests were used for qualitative data; Student’s t-test and Wilcoxon Mann–Whitney test were used for quantitative data. Clinically relevant factors associated with diagnostic yield were selected for the multivariate Cox proportional hazards model to evaluate which lesion was suitable for TBB with large GS. All statistical analyses were performed using EZR version 1.38 (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (version 3.3.2; R Foundation for Statistical Computing, Vienna, Austria) [15].
Study follow-up
If the lesion was not diagnosed, patients were recommended to undergo another diagnostic procedure such as CT-guided transthoracic needle aspiration biopsy (CTNB), repeated bronchoscopy, or surgery. In the event that the patient did not require further diagnostic procedures, the lesion was followed-up for 2 years. The final diagnosis was based on pathological evaluation or clinical follow-up.